PhyX-AF - PhyX Mechanics (Image): leaderboard
Metric: Human-verified autoformalization accuracy (%; 7 PhyX mechanics problems with the diagram as image input, Lean 4 output judged valid by two expert annotators). Source: arxiv.org. Saturation forecast: Around December 2026. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 71.4 |
| 2 | Gemini 3 Pro | 28.6 |
| 3 | Gemini 2.5 Pro | 0 |
| 4 | Qwen 3 VL 235B A22B Instruct | 0 |
| 5 | Qwen 2.5 VL 72B Instruct | 0 |
Interactive version: theaggregate.ai/benchmark?slug=phyx-af-phyx-mechanics-image · How It Works · Data refreshed daily, snapshot 2026-09-25.